BMJGH 2020 002349

Original research
Combining parenting and economic
BMJ Glob Health: first published as 10.1136/bmjgh-2020-002349 on 8 July 2020. Downloaded from http://gh.bmj.com/ on October 26, 2020 by guest. Protected by copyright.
strengthening programmes to reduce
violence against children: a cluster
randomised controlled trial with
predominantly male caregivers in
rural Tanzania
Jamie Lachman ‍ ‍ ,1,2 Joyce Wamoyi,3 Thees Spreckelsen,4 Daniel Wight,2
Jane Maganga,3 Frances Gardner1
To cite: Lachman J, Abstract Conclusion Parent training may be the active ingredient in
Wamoyi J, Spreckelsen T, Introduction Parenting programmes may reduce the risk reducing maltreatment in farmer groups with majority male
et al. Combining parenting of violence against children and improve child well-being. caregivers, while agribusiness training programmes may
and economic strengthening have unintended negative consequences on children when
However, additional economic support may be necessary
programmes to reduce
in highly deprived rural communities in sub-Saharan Africa. delivered alone. Locating parenting support in existing
violence against children: a
cluster randomised controlled Furthermore, delivering programmes within farmer groups farmer groups can engage much higher proportions of
trial with predominantly may increase male caregiver recruitment and engagement. fathers than stand-alone programmes.
male caregivers in rural Methods A parallel cluster randomised controlled trial Trial registration number
Tanzania. BMJ Global Health examined the combined and separate effects of parenting ClinicalTrials.gov: NCT02633319
2020;5:e002349. doi:10.1136/ and economic strengthening programmes on reducing
bmjgh-2020-002349 violence against children aged 0–18 years in farming
communities in Tanzania (n=248 families; 63% male
Handling editor Seye Abimbola caregivers). Eight villages were randomly assigned to four Introduction
►► Additional material is conditions (2:2:2:2): (1) 12-session parenting programme
Background on maltreatment in LMICs
published online only. To view (n=60); (2) agribusiness training (n=56); (3) parenting
please visit the journal online
Child maltreatment and other childhood
and agribusiness combined (n=72); (4) control (n=60).
(http://dx.doi.org/10.1136/ adversities occur in many low- income and
Parent-report, child-report and early childhood observation
bmjgh-2020-002349).
assessments were conducted at baseline, mid-treatment middle-income countries (LMICs) at higher
and post-treatment. Primary outcomes were child rates than in high-income countries (HICs)—
maltreatment and parenting behaviour. Secondary rates that often exceed 50%.1 In Tanzania,
Received 28 January 2020
Revised 28 May 2020 outcomes included corporal punishment endorsement, a national survey found that over 70% of
Accepted 31 May 2020 parenting stress, parent/child depression, child behaviour, respondents aged between 13 and 24 years
economic well-being and child development. had experienced physical violence before
Results At post-treatment, parents and children receiving the age of 18.2 The survey also identified a
the combined interventions reported less maltreatment correlation between physical, emotional and
(parents: incidence rate ratio (IRR=0.40, 95% CI 0.24 to sexual violence against children; approxi-
0.65; children: IRR=0.40, 95% CI 0.17 to 0.92). Parents mately 80% of respondents who experienced
reported reduced endorsement of corporal punishment sexual violence also experienced physical
(Dw=−0.43, 95% CI −0.79 to 0.07) and fewer child violence as a child, and nearly all children
behaviour problems (Dw=−0.41, 95% CI −0.77 to 0.05).
who experienced physical violence also expe-
Parents in parenting-only villages reported less abuse
rienced emotional violence. Parents and
© Author(s) (or their (IRR=0.36, 95% CI 0.21 to 0.63) and fewer child behaviour
employer(s)) 2020. Re-use other adult relatives were the most commonly
problems (Dw=−0.47, 95% CI −0.84 to 0.11). Parents in
permitted under CC BY. agribusiness-only villages reported fewer child behaviour reported perpetrators, with corporal punish-
Published by BMJ. ment considered a norm perpetuated by
problems (Dw=−0.43, 95% CI −0.77 to 0.08) and greater
For numbered affiliations see household wealth (Dw=0.57, 95% CI 0.08 to 1.06). other cultural values related to gender, family
end of article. privacy, male honour and expectations of
However, children in agribusiness-only villages reported
Correspondence to increased physical abuse (IRR=2.26, 95% CI 1.00 to 5.12) child respect and obedience.2
Dr Jamie Lachman; and less positive parenting (Dw=−0.50, 95% CI −0.91 to Recent studies have highlighted the long-
jamie.lachman@s pi.ox.a c.uk 0.10). There were no other adverse effects. term and far- reaching consequences of
Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349 1

BMJ Global Health
risks, both in high and middle- income countries.8 A

Significance of this study
systematic review demonstrated that parenting prac-
What is already known? tices in sub-Saharan Africa are associated with the same
►► Over 1 billion children experience violence each year with a dis- pattern of child outcomes as in HICs.9 There is also prom-
proportionate number in low-income and middle-income countries ising evidence that parenting programmes can effectively
(LMICs). reduce child maltreatment in LMICs.10
►► Systematic reviews suggest that parenting interventions may re- Despite emerging evidence of the effectiveness of
duce violence against children in LMICs, but that families living in parenting interventions, local governments and service
highly deprived, underserved low-income communities may require providers in LMICs face multiple challenges imple-
male participation. menting such programmes.11 Transported parenting
►► Reviews also suggest that while economic strengthening interven- programmes may not fit local contexts and may require
tions may reduce household poverty, there is limited evidence on
cultural adaptation to be relevant to local families.
their impact on reducing violence against children, with some evi-
Parents, especially fathers/male caregivers, may require
dence suggesting potential harm.
specific targeting to overcome barriers to participa-
What are the new findings? tion, including identifying entry points for programme
►► The parenting intervention delivered to farmer groups was effective delivery that harness existing social groups.12 Addi-
at reducing violence against children, with or without an economic tional components may also be necessary to address the
strengthening component. consequences of poverty, particularly relevant in highly
►► A high percentage of male caregivers were recruited to the par-
deprived rural communities in sub-Saharan Africa that
enting programme when delivered through existing farmer groups.
largely rely on subsistence agriculture with little access to
►► In villages that only received an agricultural intervention, children
reported increased physical abuse and reduced positive parenting. economic development.
What do the new findings imply? Economic strengthening programmes

►► Locating parenting interventions within existing farmer groups may Poverty has been identified as a major risk factor for phys-
increase recruitment and engagement of fathers and other male ical and emotional child abuse.13 Consequently, economic
caregivers.
strengthening programmes have been highlighted as an
►► Parenting interventions delivered to mixed-gender groups with a
important component in the reduction of violence against
large proportion of male caregivers may be effective at reducing
violence against children in highly deprived rural communities in children, particularly in multiply deprived and impover-
LMICs. ished families.14 While there is some evidence that these
►► While agricultural economic strengthening programmes may in- programmes may reduce risks of abuse and exploita-
crease economic security, they should be combined with parent- tion, the evidence of their effectiveness when delivered
ing programmes in order to reduce the risk of additional harm to alone is limited.15 Furthermore, economic strengthening
children. programmes may have unintended harmful effects on
children; a systematic review found that at least one nega-
tive child outcome in 20% of the 46 identified trials.16
violence against children, including serious physical and At the same time, there is emerging empirical evidence
mental health problems later in life, as well as difficulties suggesting the benefits of an integrated approach that
in school, employment and relationships.3 Child maltreat- combines both parenting and economic strengthening
ment can have substantial intergenerational effects, programmes.16 For instance, a recent trial in Burkina
maltreated children being more likely to maltreat their Faso found reduced harsh parenting and abuse in
own children.4 Child maltreatment is also a risk factor villages where women received an economic support
for later intimate partner violence, criminal activity, HIV package combined with a family-based intervention in
infection, transactional sex and other negative health and comparison to those which only received the economic
mental health outcomes.3 5 Furthermore, child maltreat- intervention.17 However, this study did not examine
ment has an economic impact relating to the treatment whether delivering the family-based intervention alone
of victims’ health problems, criminal justice and welfare was sufficient to reduce child maltreatment and improve
costs, and lower economic productivity.6 parent–child relationships. A recent individual partici-
pant data meta-analysis of parenting programme effec-
Parenting programmes to prevent child maltreatment tiveness in HICs found no differential effects by family
Parenting programmes have shown particular promise socioeconomic status.18
in preventing child maltreatment and other childhood The Skilful Parenting and Agribusiness Child
adversities.7 8 Parenting programmes typically aim to Abuse Prevention Study aimed to investigate the
strengthen caregiver–child relationships through positive inter-
relationship between economic strengthening
parenting and to help parents to manage child behav- programmes and parent management training in
iour problems through effective, age-appropriate, non- reducing child maltreatment. This is of particular impor-
violent discipline strategies. A meta-analysis reported that tance given the need for combined interventions that
parenting programmes successfully reduced substanti- address multiple Sustainable Development Goals (SDGs)
ated and self-reported child maltreatment and associated across economic, social and environmental domains.19
2 Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349

BMJ Global Health
Combining parenting and economic strengthening may we conducted observational assessments on early child
have a positive impact beyond SDG targets 16.2 (ie, elimi- development in participating families where there was
nating violence against children) and 5.2 (ie, eliminating a child aged 1–36 months, using a random number
violence against women and girls). Additional effects on generator to avoid selection bias by parents (n=134).
SDG targets may include ending poverty (1.1, 1.2, 1.4. Adults and children were excluded if they exhibited
1.5) and hunger (2.1, 2.2, 2.3, 2.4, 2.A), and sustaining acute mental health problems (none met this crite-
income growth (10.1) through increased agricultural rion). Families received a bar of soap after each assess-
production and food security. Therefore, we hypothe- ment as compensation. Throughout this paper ‘parent’
sised that both approaches would reduce child maltreat- is intended to include adult caregivers of children who
ment, but a combined intervention would have more are not biological parents.
robust effects across multiple outcomes associated with
increased risk of violence against children. Interventions: skilful parenting and agribusiness
We examined the differential and combined effects
Methods of two community- based interventions—a parenting
Study design and family budgeting programme and an agribusiness
We used a cluster randomised controlled trial (RCT) to training programme—with the overall goal to reduce
examine the combined and separate effects of parenting child maltreatment by improving parenting behav-
and economic strengthening programmes on reducing iours and reducing family stress due to food insecurity
child maltreatment in rural Tanzania (Pan African Clin- and financial hardship (figure 1). Skilful parenting is
ical Trials: PACTR201610001267268, preregistered 14 a 12-session group-based programme consisting of five
September 2015; C linicalTrials.gov: NCT02633319, regis- sessions on parenting skills, two on child protection and
tered at end baseline data collection on 14 December five on family budgeting. Originally developed in Kenya
2015). Reporting follows the CONSORT extension for and adapted for Tanzania, it was delivered to farmer
cluster RCTs (online supplementary tables 6 and 7). groups by Kiswahili-speaking professional trainers from
Investing in Children and Societies, an international
Setting non-profit organisation (www.icsafrica-sp.org) with local
The study was conducted in the Shinyanga Rural District, offices in Shinyanga. The agribusiness programme
populated primarily by Sukuma-speaking communities targets food and income insecurity by providing small-
mainly dependent on subsistence-level agriculture. Eight holder farmer groups access to drought-resistant seeds,
villages geographically isolated from each other by at credit for farm inputs, advice to improve farming tech-
least 20 km were selected in collaboration with the Tanza- niques and market connections. The intervention is
nian Ministry of Agriculture, Food Security, and Coop- delivered to farmer group members by trained staff from
eratives. Existing village-
based, government- organised a local economic enterprise initiative working in collabo-
farmer groups served as delivery platforms for both the ration with the Ministry of Agriculture over three inten-
parenting and agribusiness interventions. Two farmer sive workshops during the planting season and ongoing
groups were selected within each participating village support through harvesting season.
using a random selection procedure at public commu-
nity meetings in which the research investigators selected
different-coloured beads from a concealed box in the Randomisation and masking
presence of farmer group leaders (n=16 farmer groups). Cluster randomisation was conducted at village level
prior to baseline assessments, to reduce the likelihood
Participants of contamination between arms. An external researcher
Eligible parents and caregivers (n=248) were recruited used concealed computer-generated codes to randomly
from families who were members of the selected farmer allocate eight villages into three treatment arms and a
groups in each village (8–22 parents per farmer group). control arm (two villages per arm): (1) agribusiness
Inclusion criteria were: (1) age 18 or older, (2) primary plus parenting (n=72), (2) agribusiness- only (n=56),
caregiver of a child in the household aged 3–17, (3) (3) parenting-only (n=60) and (4) control (n=60). The
lived in household ≥4 nights per week, (4) registered implementing partner notified the participating families
member of agricultural farmer group and (5) provided of their allocation status after baseline data collection.
consent to participate. We recruited child respondents The implementing partner also committed to providing
aged 10–17 years from all participating families (n=176). the agribusiness, parenting or both interventions to all
If there was more than one child aged 10–17 in the villages after the end of the study. The allocation status of
household, the participating adult was asked to identify other participating villages was concealed from selected
the child with whom they were having the most difficult villages, thus reducing the potential for intervillage rival-
relationship. Inclusion criteria for child respondents ries. Research assistants conducting assessments were
included: (1) aged 10–17 years, (2) lived in the house only blind to allocation at baseline; however, the data
≥4 nights per week, (3) caregiver participating in the analyst was blind to allocation until after initial intention-
study and (4) adult and child provided consent. Finally, to-treat analyses.

BMJ Global Health
Figure 1 Theory of change model for combined skilful parenting plus agribusiness training interventions approximately here.
Power calculations be reported in separate publications. The mid-treatment

Due to funding constraints and the need to use village- data collected (June 2016) is not used or reported on in
level cluster randomisation, the sample size was limited. this paper.
Post-hoc calculations suggested 7% power to detect a
large, clustering-adjusted Dw=0.7, effect (with alpha=0.05, Outcomes
m=2 cluster per condition, n=60 participants per condi- Primary outcomes
tion, intracluster correlation=0.1).20 Primary outcomes were parent- report and child-
report of child maltreatment and parenting behaviour.
Procedure Frequency of overall child maltreatment, as well as phys-
Data collection occurred at three stages: (1) baseline ical, emotional and sexual abuse, was measured using
(September 2015), (2) mid-treatment after the parenting the ISPCAN Child Abuse Screening Tool- Trial (adult-
and child protection modules of the parenting interven- report: 22 items; child-report: 24 items).21 The Alabama
tion and before the harvesting season in the agribusiness Parenting Questionnaire (adult-report: 37 items; child-
intervention (June 2016) and (3) at post-treatment, 1 year report: 42 items) assessed parent involvement, positive
from baseline (September 2016). Because the pattern of interaction, poor supervision, inconsistent discipline and
family life varies considerably with the cycle of planting effective discipline (parent-report and child-report), as
and harvesting, it was important that both baseline and well as harsh discipline (child-report only). Items were
post-treatment assessments were conducted at the same summed for each subscale as well as a total positive
time after harvesting. Data included parent- reported parenting score by reversing negative items.
assessments for all families, child-reported assessments
for families with children aged 10–17 and early child- Secondary outcomes
hood assessments for families with children aged 0–3. Secondary outcomes included parent-report of parenting
Questionnaires were translated into Kiswahili, and back- stress (Parenting Stress Scale, 18 items), parent depres-
translated. sion (Centre for Epidemiologic Studies Depression Scale,
Trained research assistants used e- tablets to admin- 20 items) and exposure to intimate partner violence
ister consent forms and questionnaires. In addition, (Revised Conflict Tactics Scale Short Form, 10 items). We
we used audio computer-assisted self-interviewing tech- also measured parent-report and child-report of attitudes
nology to decrease stigma and increase responsivity for supporting corporal punishment (UNICEF Multiple
sensitive items regarding child maltreatment and inti- Indicator Cluster Survey, 1 item), parent and child use
mate partner violence. Research assistants were also of alcohol (1 item each) and child behaviour problems
able to explain questions in Sukuma for families with and prosocial behaviour (Strengths and Difficulties
poor understanding of Kiswahili. Post-treatment quali- Questionnaire, 25 items). Five items assessing heshima,
tative interviews and focus groups were also conducted or respectful behaviour, were added to the prosocial
to assess programme delivery and engagement; these will scale as a culturally sensitive assessment of positive child

BMJ Global Health
behaviour. Child- reported depressive symptoms were participants, we engaged with village and farmer group
measured using the 10-item Child Depression Inventory leaders at key points during the study to inform them of
short form (10 items), and child-reported sexual behav- the process and outcomes.
iour in the past month was based on items from the South
African National Survey of HIV and Risk Behaviour (15
items). Economic outcomes included household hunger Results
(Household Hunger Scale; parent- report and child- Study retention was considerably higher than antic-
report), child food consumption in the previous week ipated with 94.8% adults (n=235/248), 87.5% chil-
(child-
report), household assets (parent- report) and dren (n=154/176) and 87.8% in their early childhood
basic child necessities (child-report). Finally, early child- (n=122/139) assessments completed at post-treatment,
hood outcomes for children aged 0–36 months included with no differences between arms (figure 2). Due to delays
observational assessments of stimulation and responsive- in the implementation of the parenting intervention as a
ness (HOME Inventory), child development (Ages and result of inaccessibility of villages during the rainy season,
Stages Questionnaire) and infant/toddler growth based the final session on family budgeting was not delivered.
on measurements of weight and height. Enrolment in parenting groups was 91% of those who
In addition to the outcomes prespecified in our trial completed baseline assessments, with an average attend-
protocol, we also examined infant/child nutritional ance rate of 60% (7.5 out of 11 delivered sessions), with
health based on measurements of mid-upper arm circum- no significant differences in attendance between arms.
ference, parents’ attitudes towards family budgeting, The enrolment rate for agribusiness training was 85%,
and agricultural production based on the number of with an average attendance rate of two out of three total
90 kg bags of maize from the previous harvest (parent- sessions. Parents in the combined intervention villages
report only). These were analysed as post-hoc secondary attended the agribusiness training sessions signifi-
outcomes. cantly less than those in the agribusiness-only villages
All self-
report outcomes using frequency ratings (combined: M=1.80 sessions, SD 0.73; agribusiness-only:
referred to the past month unless otherwise specified M=2.27, SD 0.69, t=−3.46, p=0.001).
(online supplementary tables 1 and 2). Approximately two- thirds of adult participants were
fathers or other male caregivers (63% total male care-
Statistical analysis givers: 50% fathers, 6% grandfathers and 7% other rela-
This paper reports on primary and secondary outcomes tionship), primarily due to men being identified as the
at post-treatment using an intention- to-
treat analyses. family representative to farmer groups. Children were
Linear random effects regression models were used to 42% female, and 73% lived with at least one biolog-
estimate treatment effects accounting for the nesting of ical parent, 16% with either a grandparent or great-
participants in farmer groups and baseline differences grandparent and 11% with an alternative caregiver (eg,
between groups on primary and secondary outcomes. We step-parent, aunt, brother). Despite data collection being
did not adjust for differences between groups on demo- just after harvest, generally the most affluent time of
graphic characteristics due to insufficient power. Stand- year, 53% of adults reported having borrowed money to
ardised effect sizes (Dw),20 and 95% CIs were obtained meet basic needs in the past month, and 29% reported
by dividing raw coefficients for each contrast by farmer consuming two or less meals per day. Twenty-six per cent
group cluster. Incidence rate ratios (IRRs) and 95% of adults reported not having completed primary educa-
CIs were estimated with a Poisson regression for count tion, and 18% were illiterate. Sixty per cent of the chil-
outcomes. To handle missing data we use multiple impu- dren were currently enrolled in school and 52% could
tation with chained equations (R-package: Mice, V.3.6). not read or had difficulty reading. More than 70% of the
Analyses were undertaken in R V.3.3.0. Borderline effect parents reported having experienced some form of phys-
sizes of Dw>0.40 or 95% CIs within ±0.10 of 0 were consid- ical abuse as a child (table 1).
ered clinically relevant. The analysis syntax is available at Linear models and χ2 analyses examining equivalence
https://osf.io/54r9p/ among allocation arms at baseline found fewer female
caregivers (χ2(df)=26.55(3), p<0.001) and fewer female
Participant and public involvement statement children aged 0–36 months (χ2(df)=11.63(3), p<0.01) in
Representatives from the implementing organisation, the agribusiness-only villages and a higher proportion of
Investing in Children and Our Societies, were closely adults who had high scores on the discipline outcome
involved in developing and refining the research ques- (χ2(df)=24.61(12), p=0.0016) in the control villages
tions, study design and ethical procedures. We met on a compared with the other arms.
quarterly basis during the study to discuss ongoing imple- At baseline, compared with the combined interven-
mentation and emerging issues. At the end of the study, tion, parents in the parenting intervention villages
the implementing organisation and other stakeholders reported lower emotional (b=−2.42, p=0.003) and phys-
including the Government of Tanzania commented ical abuse (b=−1.99, p<0.001), and overall maltreatment
on the findings and contributed to the dissemination (b=−3.81, p=0.010). Similarly parents in agribusiness-
of results. Although it was difficult to involve study only villages reported lower agricultural assets (b=−7.746,

Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349

BMJ Global Health
Figure 2 Trial profile .
6
BMJ Global Health
Table 1 Characteristics of the sample at baseline
Total A+P A only P only Control
Age of participants, M (SD)
Adult* 43.12 (12.71) 41.65 (10.47) 44.93 (13.25) 40.64 (12.76) 44.61 (13.75)
Teen† 13.41 (2.01) 13.76 (1.87) 13.04 (2.18) 13.29 (1.81) 13.62 (2.09)
Child‡ 11.18 (3.91) 12.05 (3.82) 10.60 (4.00) 11.23 (3.59) 10.95 (4.11)
Infant (months)§ 17.94 (9.84) 19.11 (8.08) 18.44 (9.62) 18.46 (10.34) 15.28 (11.40)
Sex of participants, male, n (%)
Adult*** 157 (63.3) 33 (55.0) 63 (87.5) 27 (48.2) 34 (56.7)
Teen 104 (60.5) 31 (68.9) 32 (64.0) 18 (47.4) 23 (59.0)
Child 145 (58.5) 37 (61.7) 40 (55.6) 32 (57.1) 36 (60.0)
Infant** 71 (51.4) 23 (60.3) 13 (40.6) 14 (35.9) 21 (72.4)
Parent–child relationships: biological child, n (%)
Teen 125 (72.7) 37 (82.2) 36 (72.0) 27 (71.1) 25 (64.1)
Child 180 (72.6) 49 (81.7) 50 (69.4) 40 (71.4) 41 (68.3)
Infant 98 (71.0) 30 (78.9) 22 (68.8) 28 (71.8) 18 (62.1)
Number of people per household, M 7.30 (3.42) 7.88 (3.16) 7.49 (4.61) 6.96 (2.86) 7.30 (3.43)
(SD)
Adult marital status: married, n (%) 212 (85.5) 48 (80.0) 66 (91.7) 50 (89.3) 48 (80.0)
Adult not completed primary school, n 65 (26.2) 24 (40.0) 14 (19.4) 13 (23.2) 14 (23.3)
(%)
Adult cannot read, or reads with 61 (24.6) 23 (38.4) 10 (13.9) 11 (25.0) 14 (23.3)
difficulty, n (%)
Teen enrolled in school, n (%) 105 (61.0) 22 (48.9) 29 (58.0) 27 (71.1) 27 (69.2)
Child enrolled in school, n (%) 146 (58.9) 31 (51.7) 39 (54.2) 37 (66.1) 39 (65.0)
Parent experienced physical abuse as a 176 (71.3) 44 (73.3) 61 (84.7) 46 (82.1) 49 (83.1)
child, n (%)
*Based on number of adult participants (n=248).
†Based on number of child respondents between ages 10 and 18 (n=176).
‡Based on total number of children reported by adults (n=248).
§Based on number of children aged 0–36 months (n=134); **p<.01, ***p<.001 indicating significant differences between arms based on χ2
tests.
p=0.033). The control arm had lower levels of physical IRR=0.40, 95% CI 0.24 to 0.65; parenting-only: IRR=0.36,
abuse (b=−1.47, p=0.034) compared with the combined 95% CI 0.21 to 0.63; agribusiness-only: IRR=0.60, 95% CI
intervention arm. Children in agribusiness-only villages 0.37 to 0.95). Children in the combined villages also
reported higher basic necessity scale scores (b=1.29, reported reductions in overall child maltreatment
p=0.017) and higher positive parenting outcomes (IRR=0.40, 95% CI 0.17 to 0.92), whereas there were no
(b=7.27, p=0.028). Similarly, children in control villages differences for those who received either parenting or
reported higher basic necessity scores (b=1.05, p=0.049) agribusiness training alone (parenting- only: IRR=0.78,
and positive parenting outcomes (b=7.15, p=0.030). All 95% CI 0.36 to 1.68; agribusiness-only: IRR=1.29, 95% CI
other demographic characteristics and outcomes were 0.62 to 2.70).
equivalent at baseline. Parents in the parenting intervention villages also
reported substantially reduced emotional abuse, though
Intention-to-treat results
not those who received only the agribusiness training
Intention-to-treat results comparing each of the inter-
(combined: IRR=0.32, 95% CI 0.18 to 0.57; parenting-
vention arms with controls at post-treatment including
means and SD by allocation arm for primary outcomes only: IRR=0.27, 95% CI 0.13 to 0.59; agribusiness-only:
are summarised in tables 2 and 3, and for secondary IRR=0.57, 95% CI 0.34 to 1.19). Children also reported
outcomes in online supplementary tables 3–5. reduced emotional abuse in the combined villages
(IRR=0.27, 95% CI 0.08 to 0.89) with no child-reported
Primary outcomes differences in villages that received either intervention
Parents reported significant reductions in overall child component alone (parenting-only: IRR=0.92, 95% CI 0.30
maltreatment in all three intervention groups (combined: to 2.88; agribusiness-only: IRR=1.05, 95% CI 0.32 to 3.40).

BMJ Global Health
Table 2 Adult-report and child-report of child maltreatment outcomes based on the ISPCAN Child Abuse Screening Tool-
Intervention (ICAST-I) using an intention-to-treat analysis and adjusting for differences at baseline
Parent-report (n=248) Child-report (n=176)
Pre Post Pre Post
M (SD) M (SD) IRR* 95% CI ICC M (SD) M (SD) IRR* 95% CI ICC
Overall child maltreatment 0.135 0.365
A+P 7.19 (9.93) 2.19 (3.91) 0.40 0.24 to 0.65 12.93 (20.92) 5.20 (8.99) 0.40 0.17 to 0.92
A-only 3.38 (5.30) 2.71 (5.55) 0.60 0.37 to 0.95 11.38 (20.75) 15.45 (29.83) 1.29 0.62 to 2.70
P-only 6.43 (10.56) 1.87 (3.89) 0.36 0.21 to 0.63 12.82 (24.01) 10.03 (19.61) 0.78 0.36 to 1.68
Controls 4.63 (5.99) 4.79 (9.55) 5.62 (10.62) 8.52 (13.14)
Physical abuse 0.126 0.241
A+P 3.22 (4.94) 0.74 (1.74) 0.32 0.17 to 0.62 3.80 (7.56) 1.31 (3.29) 0.50 0.19 to 1.30
A-only 1.24 (2.18) 1.00 (2.24) 0.52 0.28 to 1.00 2.94 (6.94) 5.39 (12.34) 2.26 1.00 to 5.12
P-only 2.88 (4.48) 0.77 (1.62) 0.34 0.18 to 0.66 4.55 (9.64) 2.21 (4.14) 0.99 0.34 to 2.35
Controls 1.75 (2.93) 2.04 (4.08) 1.90 (5.50) 2.36 (5.00)
Emotional abuse 0.083 0.386
A+P 3.53 (5.46) 0.71 (1.69) 0.32 0.18 to 0.57 4.20 (8.30) 1.51 (3.46) 0.27 0.08 to 0.89
A-only 1.11 (1.78) 0.96 (1.84) 0.57 0.34 to 1.19 4.44 (7.81) 4.98 (11.50) 1.05 0.32 to 3.40
P-only 2.75 (6.17) 0.52 (1.52) 0.27 0.13 to 0.59 4.50 (10.20) 4.88 (10.64) 0.92 0.30 to 2.88
Controls 2.32 (4.02) 2.00 (5.15) 2.51 (5.46) 2.42 (4.49)
Sexual abuse 0.169 †
A+P 0.08 (0.33) 0.02 (0.13) 3.69 0.33 to 40.85 0.03 (0.16) 0.26 (0.58) 0.66 0.25 to 1.76
A-only 0.10 (0.42) 0.06 (0.29) 2.93 0.28 to 32.03 0.20 (0.53) 0.20 (0.55) 0.67 0.24 to 1.88
P-only 0.07 (0.32) 0.13 (0.44) 5.85 0.51 to 66.48 0.24 (0.59) 0.15 (0.50) 0.46 0.14 to 1.52
Controls 0.19 (0.55) 0.10 (0.41) 0.20 (0.50) 0.24 (0.61)
Neglect 0.210 0.248
A+P 0.48 (0.95) 0.74 (1.83) 1.21 0.51 to 2.84 4.93 (7.43) 2.38 (4.54) 0.44 0.18 to 1.08
A-only 1.03 (2.50) 0.75 (2.32) 0.72 0.31 to 1.71 4.00 (8.06) 5.09 (9.21) 1.31 0.58 to 2.92
P-only 0.80 (1.96) 0.58 (1.51) 0.68 0.28 to 1.65 3.76 (7.21) 2.94 (5.61) 0.56 0.23 to 1.36
Controls 0.57 (1.43) 0.75 (2.06) 1.21 (2.38) 3.74 (7.78)
*IRR for child maltreatment outcome based on comparisons of postassessment scores for each treatment arm with the control arm using Poisson
regression analyses and controlling for baseline differences. Significant effects are in bold (95% CI not crossing 1.00); borderline effects are in italics
(95% CI crossover within ±0.10 of 1.00).
†No reliable ICC estimate possible.
A-only, agribusiness only; A+P, agribusiness plus parenting; ICC, intraclass correlation coefficient; IRR, incidence risk ratio; P-only, parenting only.
There were parent- reported reductions of physical to 0.32) or children (combined: Dw=−0.05, 95% CI
abuse in all three intervention groups (combined: −0.49 to 0.38; parenting-only: Dw=−0.08, 95% CI −0.50
IRR=0.32, 95% CI 0.17 to 0.62; parenting-only: IRR=0.34, to 0.34).
95% CI 0.18 to 0.66; agribusiness-only: IRR=0.52, 95% CI Besides borderline effects for child-reported reduced
0.28 to 1.00). However, children in the agribusiness-only child neglect reported in the combined villages
villages reported substantially increased physical abuse (IRR=0.44, 95% CI 0.18 to 1.08), there were no other
(IRR=2.26, 95% CI 1.00 to 5.12). There were no differ- differences reported by parents (combined: IRR=1.21,
ences for child-reported physical abuse in the parenting 95% CI 0.51 to 2.84; parenting-only: IRR=0.68, 95% CI
intervention villages (combined: IRR=0.50, 95% CI 0.19 0.28 to 1.65; agribusiness-only: IRR=0.72, 95% CI 0.31 to
to 1.30; parenting-only: IRR=0.99, 95% CI 0.34 to 2.35). 1.71) or children (parenting-only: IRR=0.56, 95% CI 0.23
Children in the agribusiness-only villages also reported to 1.36; agribusiness-only: IRR=1.31, 95% CI 0.58 to 2.92).
reduced positive parenting, (Dw=−0.50, 95% CI −0.91 to There were no differences for sexual abuse reported
0.10), while there were borderline increases in positive by parents (combined: IRR=3.69, 95% CI 0.33 to
parenting reported by parents receiving both inter- 40.85; parenting-only: IRR=5.85, 95% CI 0.51 to 66.48;
vention components (Dw=0.26, 95% CI −0.10 to 0.62). agribusiness-only: IRR=2.93, 95% CI 0.28 to 32.03)
There were no other differences for positive parenting or children (combined: IRR=0.66, 95% CI 0.25 to
reported by parents (parenting-only: Dw=0.20, 95% CI 1.76; parenting- only: IRR=0.46, 95% CI 0.14 to 1.52;
−0.17 to 0.56; agribusiness-only: Dw=−0.03, 95% CI −0.37 agribusiness-only: IRR=0.67, 95% CI 0.24 to 1.88).

Table 3 Adult-report and child-report of parenting outcomes using an intention-to-treat analysis and adjusting for differences at baseline*
Parent-report (n=248) Child-report (n=176)
Pre Post ffect Pre Post Effect
M (SD) M (SD) Size† 95% CI M (SD) M (SD) Size† 95% CI
Total positive parenting Total positive parenting
A+P 78.07 (16.43) 84.81 (14.93) 0.26 −0.10 to 0.62 A+P 86.16 (15.45) 96.20 (14.21) −0.05 −0.49 to 0.38
A-only 78.10 (16.10) 79.99 (15.63) −0.03 −0.37 to 0.32 A-only 87.58 (15.35) 88.39 (17.82) −0.50 −0.91 to −0.10
P-only 75.68 (14.81) 82.44 (18.90) 0.20 −0.17 to 0.56 P-only 93.42 (14.70) 98.44 (11.69) −0.08 −0.50 to 0.34
Controls 75.87 (13.33) 79.34 (17.65) Controls 93.31 (13.83) 99.10 (15.30)
Positive involvement Positive involvement
A+P 15.17 (7.56) 18.74 (8.58) 0.33 −0.03 to 0.69 A+P 13.99 (7.73) 17.00 (8.71) −0.18 −0.70 to 0.34
A-only 15.11 (8.50) 15.91 (9.32) 0.03 −0.32 to 0.37 A-only 15.90 (7.39) 15.15 (10.42) −0.47 −0.95 to 0.01
P-only 14.20 (9.21) 18.01 (11.0) 0.29 −0.08 to 0.65 P-only 16.66 (8.36) 19.78 (7.38) 0.01 −0.49 to 0.51
Positive interaction Positive interaction
A+P 11.98 (5.22) 13.84 (5.38) 0.31 −0.05 to 0.67 A+P 10.85 (5.31) 13.43 (6.52) −0.00 −0.52 to 0.51
A-only 12.32 (5.01) 12.32 (5.32) 0.04 −0.30 to 0.38 A-only 11.92 (5.07) 11.83 (7.23) −0.32 −0.79 to 0.15
P-only 11.71 (5.71) 12.22 (6.46) 0.04 −0.33 to 0.40 P-only 12.48 (5.04) 14.60 (4.89) 0.10 −0.39 to 0.59
Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349

Poor supervision Poor supervision
A+P 9.47 (5.89) 8.37 (4.64) −0.22 −0.57 to 0.14 A+P 1.37 (1.17) 1.00 (1.08) −0.15 −0.63 to 0.34
A-only 9.79 (5.52) 9.01 (5.06) −0.10 −0.44 to 0.24 A-only 1.08 (1.18) 1.01 (1.34) −0.11 −0.56 to 0.35
P-only 10.64 (5.64) 8.59 (4.85) −0.23 −0.59 to 0.14 P-only 0.79 (1.06) 0.91 (1.31) −0.18 −0.65 to 0.29
Inconsistent discipline Inconsistent discipline
A+P 7.55 (3.04) 7.22 (3.80) 0.30 −0.24 to 0.83 A+P 7.58 (2.89) 7.90 (2.93) −0.13 −0.71 to 0.44
A-only 7.49 (2.98) 6.75 (3.95) 0.21 −0.31 to 0.72 A-only 9.02 (3.06) 7.84 (4.13) −0.08 −0.61 to 0.44
P-only 7.82 (4.18) 6.69 (3.43) 0.18 −0.35 to 0.71 P-only 7.92 (3.11) 8.31 (3.38) −0.02 −0.56 to 0.53
Effective discipline Effective discipline
A+P 3.93 (2.38) 3.76 (2.04) 0.11 −0.25 to 0.47 A+P 3.92 (2.81) 3.13 (2.46) −0.38 −0.88 to 0.12
A-only 3.94 (2.26) 3.70 (2.50) 0.08 −0.26 to 0.43 A-only 4.14 (3.34) 3.78 (3.95) −0.17 −0.62 to 0.29
P-only 4.23 (3.40) 3.66 (2.29) 0.07 −0.29 to 0.44 P-only 3.80 (2.74) 3.90 (2.65) −0.14 −0.62 to 0.33
BMJ Global Health
Continued
9
BMJ Global Health
Secondary outcomes
Parents in the combined villages reported less endorse-
†Hedges Dw effect sizes for parenting behaviour based on comparisons of postassessment scores for each treatment arm with the control arm using linear-mixed effects models with REML
ment of corporal punishment (Dw=−0.43, 95% CI −0.79
−0.82 to 0.45
−0.23 to 0.93
−0.50 to 0.69
to 0.07). There were also significant reductions in
parent-reported child behaviour problems in all of the
95% CI
intervention arms (combined: Dw=−0.41, 95% CI −0.77
to 0.05; parenting-only: Dw=−0.47, 95% CI −0.84 to 0.11;
‡Harsh parenting subscale was child-report only; ; Significant effects are in bold (95% CI not crossing 0); borderline effects are in italics (95% CI crossover within ±0.10 of 0).
agribusiness-only: Dw=−0.43, 95% CI −0.77 to 0.08).
There were also significant increases in household assets
Effect
Size†
−0.19
0.35
0.10
reported by parents who only received the agribusiness
training (Dw=0.57, 95% CI 0.08 to 1.06) and borderline
increases in basic child necessities reported by children
whose parents received the combined intervention
2.81 (2.71)
4.36 (3.74)
3.65 (2.51)
3.11 (2.74)
(Dw=0.43, 95% CI −0.07 to 0.79).
Borderline effect sizes were found in the combined
M (SD)
villages for parent-report of reduced adult depression

Post
Child-report (n=176)
(Dw=−0.43, 95% CI −0.91 to 0.06) and increased hours of

child labour (Dw=0.40, 95% CI −0.03 to 0.82). There were
also borderline effects found in villages that received the
agribusiness component for increased parent-report of
5.23 (3.65)
5.04 (3.91)
4.41 (3.25)
3.87 (3.19)
maize production (combined: Dw=0.54, 95% CI −0.04 to

M (SD)
1.12; agribusiness-only: Dw=0.44, 95% CI −0.11 to 1.00)

A-only, agribusiness only; A+P, agribusiness plus parenting; P-only, parenting only; REML, restricted maximum likelihood .
Pre
and measurements of mid-upper arm circumference for

children aged 0–3 (combined: Dw=0.45, 95% CI −0.10 to
0.99; agribusiness-only: Dw=0.49, 95% CI −0.01 to 0.98).
There were no significant effects in the intervention
arms on parenting stress, parent/child depression, posi-
Controls
A-only
P-only
tive child behaviour, child sexual behaviour, intimate

A+P
partner violence, household hunger and child food

Harsh parenting‡
consumption, child labour, agricultural production,

family budgeting and any of the early child development
outcomes. Analyses of parent and child alcohol use in
the past month were not conducted due to extremely low
95% CI
reports of alcohol use at baseline (adults: n=4; children:

n=3) and post-test (adults: n=13; children: n=0).
Adverse effects
Size†
ffect
In addition to the adverse effects for increased physical

abuse and reduced positive parenting reported by the
subsample of children in the agribusiness-only villages,
there were 32 cases of severe abuse reported at post-
treatment assessment distributed equally across arms.
These included reports of sexual assault (ie, either rape
M (SD)
Post
or attempted rape; n=22) or more than 30 instances of

Parent-report (n=248)
physical or emotional abuse (n=10) in the past month.

In line with our ethical protocol, these cases were imme-
and controlling for baseline differences.
diately referred to appropriate health and social services.

One caregiver died prior to post-treatment assessments.
*Alabama Parenting Questionnaire.
This was unrelated to the study.

M (SD)
Pre
Table 3 Continued
Discussion
This study is the first cluster randomised trial to examine
the combined and separate effect of parenting and
economic strengthening programmes. It thus makes an
important contribution to our knowledge and practice
regarding reducing child maltreatment in LMICs. Results
suggest that parenting programmes targeted at farmer

BMJ Global Health
groups may reduce the risk of violence against children results may have been linked to marginal increases maize
in rural Tanzania. Parents in the intervention villages production as a cash crop which subsequently reduced
reported reductions in child maltreatment, with those food and financial insecurity. Interestingly, parent-report
in combined villages also reporting reduced endorse- and child- report of household wealth was different
ment of corporal punishment. It is also encouraging that in agribusiness- only and combined villages. Whereas
children in villages receiving the combined intervention parents reported increased number of general house-
reported reduced overall maltreatment, thus corrob- hold assets in the agribusiness- only villages, children
orating adult reports. These results build on recent reported marginal increases in basic child necessities
research in Burkina Faso that found reduced violence such as school supplies and clothing in the combined
against children in families, where women received both villages. This may be due to the parenting programme’s
an economic package and family coaching.17 Importantly, emphasis on parents’ responsibility towards child well-
the current study also examined the effect of parenting being and development. However, the marginal size of
without economic strengthening, allowing us to demon- these effects requires additional research to examine
strate that parenting interventions alone may be suffi- the effect of agribusiness and parenting training on
cient to reduce maltreatment, although with smaller economic strengthening, food insecurity and financial
effects. Results are particularly encouraging due to the prioritisation in families.
high proportion of fathers and other male caregivers The harmful effects of increased physical abuse and
who are rarely included in studies,12 thus indicating that reduced positive parenting reported by children (but
parenting programmes targeted at the community level not adults) in the agribusiness-only villages is particularly
through mixed-sex social structures may be effective at concerning. Possible mechanisms include reduced parent
reaching fathers. The reductions in child maltreatment engagement and responsiveness due to increased agri-
are also promising in the Tanzanian context, where cultural activity and/or increased parent–child conflict
authoritarian and harsh parenting are normative prac- due to demands on children to assist with farming activ-
tices with high rates of violence against children reported ities or caregiving of younger children. These findings
in national surveys.2
add to the growing body of research suggesting poten-
With the exception of adult- reported borderline
tial unintended consequences of economic strength-
increases in positive parenting in the combined villages,
ening programmes on child outcomes when delivered as
the lack of improvements in positive parenting suggest
a stand-alone intervention.16 For instance, an RCT of a
that participating families may not be replacing harsh
microcredit loans programme in Bosnia found increased
practices with alternative behaviours. This may have
child labour and reduced school attendance for adoles-
been due to the fact that the parenting programme was
cents aged 16–19.24 Similarly, a cluster RCT of condi-
focused primarily on building knowledge and changing
tional cash transfers in Malawi found increased burden of
attitudes rather than the active practicing of skills, a core
household responsibilities and negative effects on mental
component for increasing positive parent involvement.22
health for adolescents.25 Furthermore, female caregivers
Further research is required to examine whether or not
the reduced harsh discipline is sustainable in the absence who received only an economic intervention in Burkina
of positive replacement behaviours. Faso reported reduced quality of parent–child relation-
It is also encouraging that adults reported signifi- ships at 24 month postassessment.17 Additionally, there is
cant reductions in child behavioural difficulties in all some evidence that economic strengthening programmes
of the intervention groups and borderline reductions targeted at primarily male-led households may lead to
in adult depression for combined and agribusiness-only increased risk of negative child-level outcomes in Uganda
groups. This suggests that both parenting and economic and Sri Lanka, as was the case in this study, although only
strengthening programmes may have a positive impact when reported by children.16 Nonetheless, it is encour-
on child and adult outcomes in low- resource settings aging that these results were not evident in the villages
such as Tanzania. These results support longitudinal that also received the parenting intervention, suggesting
research indicating multiple pathways to improving child that the inclusion of parent training may mitigate poten-
behaviour and parental mental health either indirectly tial harmful effects.
via increased food security and household wealth, or The null effects on early childhood development
directly by changing parenting behaviour.23 outcomes were not surprising given that the parenting
Findings indicating positive trends for increased intervention did not specifically address strategies for
economic well-being and food security in the combined child development. This is one of the limitations of a
and agribusiness- only villages suggest the utility of universal parenting intervention for parents of children
enhanced inputs and training in more efficient farming across the entire developmental spectrum. Additional
techniques. Although statistically not significant, villages content, specifically focused on the needs of infants
that received agribusiness training showed marginal and toddlers, may be necessary to increase effective-
increases in measurements of infant and toddler growth, ness on improving the home environment in terms of
as well as marginal effects for child-reported reduced cognitive stimulation, parental responsivity and child
household hunger in the combined villages. These development.26

BMJ Global Health
There were a number of limitations to this study. alone. Furthermore, it was especially encouraging to
Assessments primarily relied on self-report data which find unusually high participation rates of male caregivers
are susceptible to social desirability reporting biases. It given widespread concern globally that fathers are often
is possible that parents reported reduced child maltreat- difficult to recruit and retain in parenting programmes.29
ment outcomes because of their increased awareness that This resulted from the implementing agency’s strategic
these were discouraged behaviours rather than because decision to nest the programme within farmer groups
of a change in behaviour. Moreover, the divergent find- in rural communities, an existing social network that
ings for reported physical abuse in the agribusiness-only allowed for mixed-sex groups with a high proportion of
intervention raise an important issue regarding the fathers.
internal validity of results. Although the lack of conver- We recommend conducting a larger cluster RCT to
gence among most of the adult- reported and child- provide more robust evidence on the effect of combining
reported outcomes may be due to insufficient power, parenting and agribusiness training on reducing violence
divergent patterns in child and adult reporting have been against children. Additional postassessments with a longer
found elsewhere in RCTs on parenting in South Africa.27 follow-up period would allow for mediation analyses to
Nonetheless, it is encouraging that there was corrobo- further understand the mechanisms driving changes in
ration between parent- and child-report on the primary primary and secondary outcomes, especially concerning
outcome of reduced child maltreatment and emotional the interaction between parenting and economic
abuse by families receiving the combined parenting and outcomes. Observational assessments of parent–child
agribusiness intervention. interaction would also provide a more robust estimation
Another limitation was low power and precision to of effects that are less susceptible to social desirability.
detect intervention effects due to the small sample size of Other potential study designs may include a factorial
villages and recruited families in the cluster randomised experimental design to provide further understanding
design. This is an issue particularly for the child (n=176) of the differential effects of specific components of
and early childhood (n=138) samples which were lower this complex family intervention. Furthermore, results
than the total sample size (n=248), thus increasing the suggest that a universal parenting programme delivered
chance of a potential Type I error in the results. It was to families with children across the entire age range from
also not possible to directly compare adult-report and 0 to 17 years may require additional early childhood
child-
report data since adults reported on any child development modules for those who have younger chil-
between the ages of 3 and 17 regardless of whether there dren. Lastly, studies examining the effects of economic
was a child between the ages of 10 and 17 who also partic- strengthening interventions delivered without additional
ipated in the study. Future studies would benefit from a parenting components should be conducted with caution
larger and more selective sample with parent–child pairs given the negative impacts reported by children in this
to allow more robust comparisons between adult-report study. At the very least, we recommend that researchers
and child-report data. Finally, results are limited by the include assessments of parenting behaviour to examine
fact that we were only able to conduct post-treatment possible harmful effects.
assessments immediately after the interventions were
delivered. Initially, the study was planned to have both Conclusion
an immediate postintervention and 4 month follow-up This study makes a valuable contribution to our knowl-
assessment. Due to delays in the inception of the study edge of the combined effect of parenting and economic
and in programme implementation, the initial imme- strengthening programmes. Results indicate that while
diate postintervention assessment was changed to a parenting and agribusiness training combined may be
mid-treatment assessment resulting in only one postinter- most effective at reducing risks of violence against chil-
vention assessment. A study with a longer-term follow-up dren on a range of outcomes, parenting programmes
would have also enabled the exploration of delayed treat- delivered alone may also be effective. Moreover, findings
ment effects. For example, other studies have reported also suggest that implementing an agricultural economic
that parents may require more time to practice new strengthening programme alone may, according to
approaches to discipline in order to implement them child reports, have some adverse effects on parent–child
with consistency.28 interaction leading to increased risk of violence against
There are a number of strengths of the study. It is children. This is concerning given the pervasiveness of
rigorous compared with existing studies that investi- economic strengthening interventions in LMICs, and the
gate parenting interventions in LMICs.10 The cluster inclusion of economic strengthening as part of recom-
randomised design with both child-reported and parent- mended strategies to end violence against children.14
reported data allowed for a robust comparison of outcome Nonetheless, these findings are particularly important
effects with reasonable estimates of causality. The trial also given the barriers to engaging fathers experienced
provided a unique opportunity to investigate both the in other settings,12 and the limited evidence available
combined impact of parenting and economic strength- regarding the effectiveness of complex community
ening interventions on reducing risks of violence against and family-based interventions in LMICs. Findings also
children, as well as the impact of these interventions provide an important foundation for future research

BMJ Global Health
examining the potential for complex interventions to participant data and analysis syntax, is available via Open Science Framework with
no end date to anyone who wishes to access the data (https://osf.io/54r9p/).
accelerate positive impacts on multiple SDG targets
across economic, social and environmental domains in Open access This is an open access article distributed in accordance with the
Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits
Africa and beyond.30 others to copy, redistribute, remix, transform and build upon this work for any
purpose, provided the original work is properly cited, a link to the licence is given,
Author affiliations and indication of whether changes were made. See: https://creativecommons.org/
1
Department of Social Policy and Intervention, University of Oxford, Oxford, UK licenses/by/4.0/.
2
MRC/CSO Social and Public Health Sciences Unit, University of Glasgow, Glasgow,
ORCID iD
UK
3 Jamie Lachman http://orcid.org/0000-0001-9475-9218
National Institute for Medical Research Mwanza Research Centre, Mwanza,
Mwanza, United Republic of Tanzania
4
School of Social and Political Sciences, University of Glasgow, Glasgow, UK
Twitter Jamie Lachman @Banjomax
References
Contributors JL is the primary author and principal investigator on the study 1 Stoltenborgh M, Bakermans-Kranenburg MJ, van Ijzendoorn MH,
along with JW. JL, FG, DW and JW designed the evaluation and secured funding. et al. Cultural-geographical differences in the occurrence of child
JL, JW and JM managed data collection, trained and supervised the survey team physical abuse? A meta-analysis of global prevalence. Int J Psychol
and conducted qualitative research themselves. JL led the analysis and writing, 2013;48:81–94.
2 UNICEF. Violence against children in Tanzania: findings from a
with statistical analyses led by TS. DW, FG, JW, TS and JM provided considerable
national survey, 2009. Dar es Salaam, Tanzania: UNICEF Tanzania,
input to the writing of the manuscript, and all authors critically reviewed the Division of Violence Prevention, National Center for Injury Prevention
penultimate draft. and Control, Centers for Disease Control and Prevention, and
Funding The Skilful Parenting and Agribusiness Child Abuse Prevention Study Muhimbili University of Health and Allied Sciences, 2011.
3 Norman RE, Byambaa M, De R, et al. The long-term health
was supported by the UBS Optimus Foundation (Grant: 7849.09), the Netherlands
consequences of child physical abuse, emotional abuse, and
Ministry of Foreign Affairs, and the Complexity in Health Improvement Programme neglect: a systematic review and meta-analysis. PLoS Med
of the Medical Research Council MRC UK (Grant: MC_UU_12017/14). None of the 2012;9:e1001349.
funding sources contributed to the writing of the manuscript or decision to submit 4 Pears KC, Capaldi DM. Intergenerational transmission of abuse:
it for publication. The corresponding author had full access to all the data in the a two-generational prospective study of an at-risk sample. Child
study and had final responsibility for the decision to submit for publication. Abuse Negl 2001;25:1439–61.
5 Cluver L, Orkin M, Boyes M, et al. Transactional sex amongst AIDS-
Competing interests The authors declare the following competing interests: JL orphaned and AIDS-affected adolescents predicted by abuse and
reports grants from UBS Optimus Foundation, grants from Netherlands Ministry of extreme poverty. J Acquir Immune Defic Syndr 2011;58:336–43.
Foreign Affairs, grants from Complexity in Health Improvement Programme of the 6 Fang X, Zheng X, Fry DA, et al. The economic burden of violence
Medical Research Council MRC UK, during the conduct of the study; grants from against children in South Africa. Int J Environ Res Public Health
UNICEF Thailand, grants from UNICEF Philippines, personal fees from Maestral 2017;14:1431.
International, grants from EU Horizon 2020, personal fees from Oxford Policy 7 Leijten P, Gardner F, Melendez-Torres G, et al. What to teach parents
to reduce disruptive child behavior: two meta-analyses of parenting
Management, personal fees from Clowns Without Borders South Africa, grants program components. Journal of the American Academy of Child &
from Oak Foundation, outside the submitted work; and JL is the codeveloper of a Adolescent Psychiatry 2018;58:180–90.
non-profit parenting program with the WHO ‘Parenting for Lifelong Health’, under a 8 Chen M, Chan KL. Effects of parenting programs on child
creative commons license, and receives fees as a trainer for this. JW reports grants maltreatment prevention: a meta-analysis. Trauma Violence Abuse
from UBS Optimus foundation during the conduct of the study. TS reports personal 2016;17:88–104.
fees from International Catholic Relief Services, personal fees from Department 9 Devlin AM, Wight D, Fenton C. Are parenting practices associated
of Education, University of Oxford, personal fees and other from Journal of Child with the same child outcomes in sub-Saharan African countries as
in high-income countries? A review and synthesis. BMJ Glob Health
and Adolescent Mental Health outside the submitted work. JM reports grants from
2018;3:e000912.
UBS Optimus Foundation, grants from Netherlands Ministry of Foreign Affairs, from 10 Gardner F, Knerr W, Lachman JM, et al. Parenting for preventing risk
Complexity in Health Improvement Programme of the Medical Research Council of child maltreatment in low- and middle-income countries: updated
MRC UK during the conduct of the study. DW reports grants from UBS Optimus systematic review and meta-analysis. forthcoming.
Foundation, grants from Netherlands Ministry of Foreign Affairs, from Complexity in 11 Parra-Cardona R, Leijten P, Lachman JM, et al. Strengthening a
Health Improvement Programme of the Medical Research Council MRC UK, during culture of prevention in low- and middle-income countries: balancing
the conduct of the study; and DW has devoted considerable time to developing scientific expectations and contextual realities. Prev Sci 2018:1–11.
a parenting programme in Uganda (Parenting for Respectability) which might be 12 Panter-Brick C, Burgess A, Eggerman M, et al. Practitioner review:
Engaging fathers--recommendations for a game change in parenting
perceived to create a bias towards interpreting parenting interventions as effective.
interventions based on a systematic review of the global evidence. J
FG reports grants from NIHR, during the conduct of the study; grants from UNICEF, Child Psychol Psychiatry 2014;55:1187–212.
grants from EU Horizon 2020, grants from UBS Optimus Foundation, grants from 13 Stith SM, Liu T, Davies LC, et al. Risk factors in child maltreatment:
John Fell Oxford University Press Research Fund, personal fees from Blueprints a meta-analytic review of the literature. Aggress Violent Behav
for Health Youth Development, University of Colorado, Boulder, personal fees 2009;14:13–29.
from Trygfonden Foundation, Denmark, outside the submitted work. FG is also a 14 World Health Organization. INSPIRE: seven strategies for ending
codeveloper of a non-profit parenting programme with WHO, ‘Parenting for Lifelong violence against children. Geneva: World Health Organization, 2016.
Health’, and receives no fees for this. 15 Marcus R, Page E. Economic strengthening activities in child
protection interventions: an adapted systematic review. London:
Patient and public involvement Patients and/or the public were involved in the ODI, 2014.
design, or conduct, or reporting, or dissemination plans of this research. Refer to 16 Ellis CM, Chaffin J. Evaluations of outcomes for children and
the Methods section for further details. youth from NGO-supported microeconomic interventions: a
research synthesis. Enterprise Development and Microfinance
Patient consent for publication Not required. 2015;26:104–21.
17 Ismayilova L, Karimli L. Harsh parenting and violence against
Ethics approval Ethical approval was received from Tanzania Lake Zone Institutional
children: a trial with ultrapoor families in Francophone West Africa. J
Review Board (3 September 2015) and Oxford Social Sciences and Humanities Clin Child Adolesc Psychol 2020;49:1–18.
Interdivisional Research Ethics Committee (SSD/CUREC1A/14-SSH_C2_15_023). 18 Gardner F, Leijten P, Harris V, et al. Equity effects of parenting
Provenance and peer review Not commissioned; externally peer reviewed. interventions for child conduct problems: a pan-European individual
participant data meta-analysis. Lancet Psychiatry 2019;6:518–27.
Data availability statement Data are available in a public, open access 19 Nilsson M, Griggs D, Visbeck M. Policy: map the interactions
repository. Data collected for this study, including de-identified individual between sustainable development goals. Nature 2016;534:320–2.

BMJ Global Health
20 Hedges LV. Effect sizes in cluster-randomized designs. Journal of 26 Black MM, Hurley KM. Early child development programmes: further
Educational and Behavioral Statistics 2007;32:341–70. evidence for action. Lancet Glob Health 2016;4:e505–6.
21 Meinck F, Boyes ME, Cluver L, et al. Adaptation and psychometric 27 Cluver LD, Meinck F, Steinert JI, et al. Parenting for lifelong
properties of the ISPCAN child abuse screening tool for use in trials health: a pragmatic cluster randomised controlled trial of a non-
(ICAST-Trial) among South African adolescents and their primary commercialised parenting programme for adolescents and their
caregivers. Child Abuse Negl 2018;82:45–58. families in South Africa. BMJ Glob Health 2018;3:e000539.
22 Kaminski JW, Valle LA, Filene JH, et al. A meta-analytic review of 28 van Aar J, Leijten P, Orobio de Castro B, et al. Sustained, fade-
components associated with parent training program effectiveness. out or sleeper effects? A systematic review and meta-analysis of
J Abnorm Child Psychol 2008;36:567–89. parenting interventions for disruptive child behavior. Clin Psychol
23 Gershoff ET, Aber JL, Raver CC, et al. Income is not enough: Rev 2017;51:153–63.
incorporating material hardship into models of income 29 Siu GE, Wight D, Seeley J, et al. Men’s Involvement in a Parenting
associations with parenting and child development. Child Dev Programme to Reduce Child Maltreatment and Gender-Based
2007;78:70–95. Violence: Formative Evaluation in Uganda. Eur J Dev Res
24 Augsburg B, De Haas R, Harmgart H, et al. The impacts of 2017;29:1017–37.
Microcredit: evidence from Bosnia and Herzegovina. Am Econ J 30 Cluver LD, Orkin FM, Campeau L, et al. Improving lives by
Appl Econ 2015;7:183–203. accelerating progress towards the un sustainable development goals
25 Baird S, McIntosh C, Özler B. Cash or condition? Evidence from a for adolescents living with HIV: a prospective cohort study. Lancet
cash transfer experiment. Q J Econ 2011;126:1709–53. Child Adolesc Health 2019;3:245–54.

BMJGH 2020 002349

Uploaded by

Copyright:

Available Formats

BMJGH 2020 002349

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

BMJGH 2020 002349

Uploaded by

Copyright:

Available Formats

Original research

Combining parenting and economic

Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349  1

risks, both in high and middle-­ income countries.8 A

What do the new findings imply? Economic strengthening programmes

2 Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349

Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349 3

Power calculations be reported in separate publications. The mid-­treatment

4 Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349

Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349 5

Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349

Figure 2 Trial profile .

Table 1 Characteristics of the sample at baseline

Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349 7

8 Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349

Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349

villages for parent-­report of reduced adult depression

(Dw=−0.43, 95% CI −0.91 to 0.06) and increased hours of

maize production (combined: Dw=0.54, 95% CI −0.04 to

1.12; agribusiness-­only: Dw=0.44, 95% CI −0.11 to 1.00)

and measurements of mid-­upper arm circumference for

tive child behaviour, child sexual behaviour, intimate

partner violence, household hunger and child food

consumption, child labour, agricultural production,

reports of alcohol use at baseline (adults: n=4; children:

n=3) and post-­test (adults: n=13; children: n=0).

In addition to the adverse effects for increased physical

or attempted rape; n=22) or more than 30 instances of

physical or emotional abuse (n=10) in the past month.

diately referred to appropriate health and social services.

This was unrelated to the study.

10 Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349

Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349 11

12 Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349

Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349 13

14 Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349

You might also like

Lachman J, et al. BMJ Global Health 2020;5:e002349. doi:10.1136/bmjgh-2020-002349 1

risks, both in high and middle- income countries.8 A

Power calculations be reported in separate publications. The mid-treatment

villages for parent-report of reduced adult depression

1.12; agribusiness-only: Dw=0.44, 95% CI −0.11 to 1.00)

and measurements of mid-upper arm circumference for

n=3) and post-test (adults: n=13; children: n=0).